NSTU-BDTAKA: NSTU Bangladeshi Paper Currency Dataset

Published: 28 October 2023| Version 1 | DOI: 10.17632/w4y6h723xg.1
Mohammad Rony,


The "NSTU-BDTAKA" dataset has been carefully organized to facilitate two primary tasks: the identification and detection of taka, with a particular emphasis on Bangladeshi paper money. A hierarchical folder structure inside the "Detect" folder divides data into "train," "test," and "validation" subsets for the Taka detection subset, with each including "images" and "labels" subfolders. Annotation files in YOLO format are kept in the "labels" folder, which guarantees accurate object localization for YOLOv5-based object identification models. The dataset is divided into "train," "test," and "validation" subsets in the taka recognition subset, with distinct class folders for different amounts of money. These class folders include images of particular denominations, which simplifies the dataset for training recognition models and allows for precise currency classification based just on image data. This meticulous structure supports the development, training, and evaluation of models, making the "NSTU-BDTAKA" dataset a valuable resource for researchers and practitioners in computer vision and currency analysis.


Steps to reproduce

To reproduce the "NSTU-BDTAKA" dataset for academic purposes, it is essential to meticulously follow a systematic set of steps: Capture high-resolution images of Bangladeshi paper currency in diverse conditions, annotate them with YOLO format for Taka detection, split into "train," "test," and "validation" subsets , apply data augmentation, preprocess images, capture additional images for Taka recognition, label them by denomination, normalize recognition images, organize the dataset into a hierarchical folder structure.


Noakhali Science and Technology University Faculty of Computer Science and Telecommunication Engineering


Object Detection, Recognition